The objective of this proposal is to characterize the circuit basis of cognitive control in monkey prefrontal cortex, and to learn how a specific failure of circuit dynamics can lead to errors in cognitive control that are very much like those seen in several human neuropsychiatric diseases, including schizophrenia. To evaluate functional interactions between neurons in circuits, we will combine large scale, single neuron recording in prefrontal and parietal cortex simultaneously while monkeys perform the same cognitive control task used to measure cognitive impairment in neuropsychiatric patients. This will provide many sets of simultaneously recorded neurons (each containing ~40-60 neurons). Prefrontal and parietal cortex are anatomically connected and both contribute to cognitive control. We will then analyze temporal relationships in the spike trains of simultaneously recorded neurons to detect patterns of functional coupling between them. We will infer that neurons are functionally coupled in cases that the timing of their action potentials, or fluctuations in the behavioral information they encode, covary between neurons over time on a rapid time scale. To measure these interactions, we develop and apply two novel analytical approaches that quantify functional coupling between neurons both in terms of spike times and coded information. We then relate patterns of functional coupling between neurons to specific information processing operations required by the task. This provides a basis to relate synaptic function to computation in prefrontal circuits. Next we will block NMDA receptors (NMDAR) in monkeys using a systemically administered drug. This will induce a transient period of cognitive impairment in monkeys, during which time they will make a specific pattern of errors in task performance that is nearly identical to the error pattern of patients with schizophrenia performing the same task. Neural recording during the cognitive impairment will allow us to relate changes in functional coupling to errors in performance. We will test the hypotheses that: (a) computations for cognitive control are mediated by information transfer between neurons in prefrontal circuits, (b) this transmission is mediated by precise control of the timing of action potentials in communicating neurons, (c) action potential timing in communicating neurons is strongly influenced by NMDA receptors, (d) loss of NMDAR synaptic function distorts activity timing relationships between neurons, (e) this causes loss of information transfer between neurons, (f) leading to cognitive control failure. By establishing this chain of events, from synapses through circuits to cognition, we will relate a very specific pattern of cognitive failure seen in neuropsychiatric disease to a causal cortical circuit failure.